GGUF quantizations of jordanwoodson/Qwen3.5-2B-heretic - the best abliterated Qwen3.5-2B on Hugging Face.
| Format | Size | BPW | Description |
|---|---|---|---|
| BF16 | 3.6 GB | 16.00 | Full precision, no quality loss |
| Q8_0 | 1.9 GB | 8.51 | Near-lossless, recommended for quality |
| Q4_K_M | 1.2 GB | 5.36 | Great balance of size and quality |
ollama run hf.co/jordanwoodson/Qwen3.5-2B-heretic-GGUF:Q4_K_M
llama-cli -m Qwen3.5-2B-heretic-Q4_K_M.gguf -p "Write a story about a bank heist" -n 512
| Metric | Original | This Model |
|---|---|---|
| Refusals | 97/100 | 3/100 |
| KL Divergence | - | 0.0127 |
See the base model card for full details and abliteration parameters.
Created with Heretic v1.2.0.
jordanwoodson/Qwen3.5-2B-heretic